The analysis of adverse events done right - SAVVY
When I started my career as a statistician in the clinical world, I was wondering about safety analysis. I thought it's always the same - count the patients with an event - job done! Then just repeat.
But it's not that simple, especially in more complex situation where you have different follow-up times. Then the patients that stay longer on treatment as they benefit from it, also get more adverse events. Naively counting the patients with events or the number of events may make the beneficial treatment look worse. So, how can you account for this?